上海单季晚粳稻主要生育性状的动态模拟

Dynamic Simulation on Main Growth Traits of Single-Cropping Late Japonica Rice in Shanghai

20世纪50年代以来,上海地区粮食耕作制度和水稻品种经过不断更替,发展为目前的稻麦(或稻油)轮作为主的一年两熟制,其中水稻主要以单季晚粳稻为主。为了对单季晚稻的生长状况进行有效预估,采用线性逐步回归方法建立水稻生育期间隔日数与气象因子的关系模型;选择已有成熟的模型(或模块),在参数定标的基础上,建立水稻茎蘖动态、叶龄和灌浆过程模型。选用2014-2016年杂交粳稻"秋优金丰"共16个播期的田间试验观测资料进行参数定标及模型有效性验证,在此基础上模拟水稻各生育性状的动态变化,并对模拟效果进行误差分析。结果表明,各模型均能较好地模拟单季晚粳稻发育期、茎蘖动态、叶龄和灌浆过程,发育期和叶龄模型模拟值与实测值的相关系数(R)均达0.95以上(P<0.001),归一化均方根误差(RMSEn)均在10以下;茎蘖动态和灌浆动态模型模拟值与实测值的R值均达0.85以上(P<0.001),RMSEn分别为19.8和31.2,粒重的模拟误差主要出现在灌浆的中后阶段。总体来看,各模型对上海地区单季晚粳稻具有较好的模拟性能,能够为生育期、茎蘖消长、叶龄和灌浆过程的动态预测提供依据。

Since the 1950 s, the food cultivation system and rice varieties in Shanghai have undergone continuous replacement, and they have developed into the current wheat-rice or canola-rice as the main annual rotation system, in which rice is mainly single-cropping late japonica rice. In order to effectively estimate the growth status of single-cropping late rice, a linear stepwise regression method was used to establish a relationship model for the length of adjacent rice growth stage and meteorological factors. Three process models such as leaf age, tillering dynamic and grain filling were established based on selection of well-established modules and parameter calibration. Observation data from 16 sowing dates field experiment of the hybrid rice ＂Qiuyou Jinfeng＂ from 2014 to 2016 were selected to verify the parameters and the validity of the model. The dynamic changes of rice＇s growing traits were simulated, and error analysis was performed. The results showed that all models performed good in simulating the development stages, tillering dynamics, leaf age and grain filling process, and the correlation coefficient（R） between the simulated values and measured values in development stages and leaf age are both larger than 0.95（P〈0.001）, the normalized root mean squared error（RMSEn） are both below 10. The R value between the simulated and measured values of the tillering dynamics and grain filling dynamic model are both larger than 0.85（P〈0.001）, and the RMSEn are 19.8 and 31.2, respectively. The simulation error of grain weight mainly occurs in the middle and late stage of grain filling. Overall, each model has good simulation performance for single cropping late rice in Shanghai, and can help for dynamic prediction of growth period, stem length, leaf age, and grouting process estimation.